healthcare benchmark
HealthBench Hard Leaderboard
HealthBench Hard is a challenging variation of HealthBench that evaluates large language models in healthcare using 5,000 multi-turn conversations and evaluation criteria validated by 262 physicians from 60 countries.
Updated Oct 23, 2026
Models11
Model coverage11
MetricScore
EvidenceC
HealthBench Hard Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelME | Score42.80% | Percentile100.00% | Participants11 | EvidenceC | Evaluated |
| Rank02 | ModelOP | Score36.20% | Percentile90.00% | Participants11 | EvidenceC | Evaluated |
| Rank03 | ModelOP | Score33.10% | Percentile80.00% | Participants11 | EvidenceC | Evaluated |
| Rank04 | ModelOP | Score32.70% | Percentile70.00% | Participants11 | EvidenceC | Evaluated |
| Rank05 | ModelOP | Score32.00% | Percentile60.00% | Participants11 | EvidenceC | Evaluated |
| Rank06 | ModelOP | Score30.00% | Percentile50.00% | Participants11 | EvidenceC | Evaluated |
| Rank07 | ModelOP | Score25.90% | Percentile40.00% | Participants11 | EvidenceC | Evaluated |
| Rank08 | ModelOP | Score22.90% | Percentile30.00% | Participants11 | EvidenceC | Evaluated |
| Rank09 | ModelBY | Score15.30% | Percentile20.00% | Participants11 | EvidenceC | Evaluated |
| Rank10 | ModelOP | Score10.80% | Percentile10.00% | Participants11 | EvidenceC | Evaluated |
| Rank11 | ModelOP | Score1.60% | Percentile0.00% | Participants11 | EvidenceC | Evaluated |
HealthBench Hard Highlights
The leading models and scores on this benchmark.
HealthBench Hard Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for HealthBench Hard
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is HealthBench Hard?
What HealthBench Hard measures and how its scores work.
HealthBench Hard is a healthcare benchmark for evaluating large language models through 5,000 multi-turn conversations.
It measures large language models' performance and safety in healthcare using a Score reported as a ratio.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of C.
- Family
- HealthBench Hard
- Modality
- text
- Primary category
- healthcare
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about HealthBench Hard.
Which model scores highest on HealthBench Hard?
Muse Spark is currently ranked first with 42.80%.
What are the top three models on HealthBench Hard?
The current leaders are Muse Spark (42.80%), GPT-6.1 Sol (36.20%), and GPT-5.6 Sol (33.10%).
Which HealthBench Hard model has the lowest official input price?
GPT-5.6 Luna has the lowest matched official input price at $0.20 input / $1.2 output per 1M tokens.
Which models are fastest among HealthBench Hard results?
The fastest matched records are GPT OSS 20B (1,000.00 tok/s via Groq), GPT OSS 120B (500.00 tok/s via Groq), and GPT-5.6 Luna (215.52 tok/s via OpenAI).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.